On 21.05.2014 14:58, Tycho Andersen wrote:
On Wed, May 21, 2014 at 11:35:32AM +0200, Niklas Semmler wrote:
Hi tycho,
thanks for your answer. By now the error has disappeared again,
probably related to my exclusion of the pacman widget. Anyway I am
still interested in having a systematic way of debugging these kinds
of errors.
So it looks (?) to me like before 409130 the pacman widget might have
blocked on actually calling pacman, which could trigger some jerkiness
(although I wouldn't think it would kill qtile entirely, unless
`checkupdates` blocked forever. Are you experiencing the issue with
the current git head and the pacman widget?
Yes, that might be the cause. I pulled a new qtile version and will
inspect if the situation changes.
Your idea of connecting a debugger sounds good. But so far I have
not been able to manage that to work. I tried attaching gdb to the
running process:
* Recompiling python with debug symbols
* Load the python gdb tools from source in gdb
* attaching gdb to the process
What gives me the following error:
Python Exception <class 'gdb.error'> There is no member named co_name.:
Error occurred in Python command: There is no member named co_name.
Unfortunately this looks to be a dead end. With my little knowledge
of gdb and python on this level, there appears to be nothing I could
do.
My best idea would be to open qtile and add a "manhole" as in this
example.
https://pypi.python.org/pypi/manhole
What are your thoughts?
I've never used manhole, but it looks interesting. Now that you
mention it, I think we've had problems attaching debuggers
(specifically pdb) in the past (most of them die when we do
gobject.threads_init()), although I thought someone had had success
more recently, I can't remember exactly what they did. Have you tried
pdb instead of gdb?
I was looking into pdb, but as far as I understand there is no way to
attach pdb to a running process. Please correct me if I am wrong.
It would be great to look into qtile when something breaks.
-- Niklas
\t
Best,
Niklas
On 16.05.2014 14:50, Tycho Andersen wrote:
On Fri, May 16, 2014 at 12:35:42AM +0200, Niklas Semmler wrote:
Error reappeared. Qtile freezes suddenly.
Battery widget & NetGraph are deactivated. Pacman widget runs.
strace doesn't show anything up just that process reads or waits.
Debug is filled with waste.
Can you post said waste? :-)
A suspend normally saves me, most of the times. It's just a big gamble.
Sounds like we are blocking somewhere, then. The question is where.
Can you attach pdb when you're in this state and potentially figure
out where things are frozen? (Or you could send it SIGINT and get a
stack trace if it crashes.)
\t
--
You received this message because you are subscribed to the Google Groups
"qtile-dev" group.
To unsubscribe from this group and stop receiving emails from it, send an email
to [email protected].
For more options, visit https://groups.google.com/d/optout.
--
You received this message because you are subscribed to the Google Groups
"qtile-dev" group.
To unsubscribe from this group and stop receiving emails from it, send an email
to [email protected].
For more options, visit https://groups.google.com/d/optout.
